Sampling bias in multiscale ant diversity responses to landscape composition in a human-disturbed rainforest
MetadataShow full item record
Recent studies have shown that several sources of variation can influence our ability to quantify biological responses to environmental variables, and that spatial scales are important in this process. For instance, sampling methods may differ in their efficiency or specificity, leading to different inferred relationships between community responses and landscape composition—i.e., forest cover (%), landscape heterogeneity, edge effects, and functional connectivity. Consequently, this can also influence the predictive power of the models when evaluating organisms as bioindicators of habitat loss and land use modification. Here, we evaluated how sampling methods (i.e., Winkler, pitfall, beating, and baits) influence our capacity to assess the scale of effect of two landscape composition metrics on ant diversity. We conducted ant sampling in 16 landscapes within a Mexican tropical rainforest and assessed the relationship between species richness and landscape composition metrics through buffers with 12 different spatial extents (from 50 to 1000 m). We found that the sampling method influenced the scale of effect when evaluating the relationships between ant species richness and forest cover and landscape heterogeneity. Combining all sampling methods, we found that the scales that best explained ant species richness were 700 m for forest cover and 900 m for landscape heterogeneity. Therefore, we highlight that our ability to detect ant-based diversity responses to environmental variables depends on the sampling method and spatial extent used in the study.